Novel framework for Zero-Shot Style Alignment in Text-to-Image generation, using minimal attention sharing and ensuring consistent style transfer without fine-tuning.
generative-model
generative
depth-image
clip
canny-edge-detection
diffusion
zero-shot-learning
diffusion-models
t2i
reference-images
controlnet
style-aligned
-
Updated
Sep 21, 2024 - Jupyter Notebook